https://orcid.org/0000-0002-0079-2488
Figures
Abstract
Sickle cell disease (SCD) affects millions of individuals worldwide, and is characterized by both acute, episodic pain and chronic, persistent pain. Despite the significant burden of the disease, clinicians continue to face significant challenges in treating SCD pain due to variability in pain experiences. The objectives of this study were (1) to identify distinct pain subgroups based on demographic and biopsychosocial characteristics and (2) to evaluate the relationship between the subgroups and pain impact – a SCD-disease specific measure of pain interference. To achieve these objectives, we performed a hierarchical cluster analysis on a cross-sectional sample of adults with SCD who are enrolled in the Globin Research Network Data and Discovery (GRNDaD) registry. Five hundred thirty-two participants (61% females and 64% with chronic pain) were included in the analysis. Six distinct subgroups were identified, 3 with chronic pain (Clusters 1–3) and 3 without chronic pain (Clusters 4–6). Despite differences in biological markers of disease severity such as genotype, hemoglobin and fetal hemoglobin percentage, chronic pain subgroups had comparable odds of reporting worse pain impact, suggesting that chronic pain has a disproportionate influence on SCD pain when compared to other factors. Longitudinal studies are needed to further validate these findings and to determine how these pain subgroups may change. Overall, our findings indicate that understanding and preventing chronic pain in SCD must be a top priority to improve the quality of life of those living with SCD.
Citation: Kenney MO, Wilson S, Rosser M, Lanzkron S, Kanter J, Padrino S, et al. (2025) Identifying distinct subgroups with severe pain in sickle cell disease: A cluster analysis of the GRNDaD multi-center registry. PLoS One 20(5): e0320889. https://doi.org/10.1371/journal.pone.0320889
Editor: Deogratias Munube,, Makerere University / Mulago National Referral Hospital, UGANDA
Received: September 13, 2024; Accepted: February 25, 2025; Published: May 15, 2025
Copyright: © 2025 Kenney et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data that supports the findings in this study contains potentially identifying or sensitive patient information. The data are available from Johns Hopkins University School of Medicine’s Institutional Review Board (jhmeirb@jhmi.edu) upon reasonable request.
Funding: MOK received funding from National Heart, Lung, and Blood Institute (Grant #5K01HL169339-02). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have read the journal’s policy and have the following competing interests: SW has consulted with Agios Pharmaceuticals and Solas BioVentures. SL is on adjudication committee for Bluebird Bio, Novo Nordisk and Beam Therapeutics, and has served as a consultant for Novartis, Merck, Glycomimetics, Pfizer. PD has honoraria from Bluebird Bio and is a advisor for NMDP. JK is a consultant for Guide point Global, GLG, Novartis, and Bluebird and has received honorarium from Novartis, Oric, Bausch, Glycomimetics, Chiesi, Bausch, GSK, and Chiesi. PD with honoraria from Vertex, Bluebird, Cheisi, and is an advisor for the NMDP. DM serves as a consultant for GBT, Pfizer, and Novartis. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Introduction
Sickle cell disease (SCD) is the most common inherited hemoglobinopathy globally. Pain is the hallmark of SCD, and is associated with high healthcare utilization and premature mortality [1]. Acute, episodic severe pain evolves into chronic/persistent pain in 50%-60% of individuals by late adolescence to early adulthood [2,3]. Chronic SCD pain contributes to poor health-related quality of life in SCD [4,5]. However, despite the significant impact of SCD pain on public health and individual well-being, effective pain therapies are few, leading to disparities in pain outcomes.
Variability in pain symptoms among individuals with SCD has contributed to poor pain outcomes. This variability, present both across different individuals and within the same individual over time, includes variations in pain frequency and distribution, and is also noted among SCD subgroups with chronic/persistent pain [5–8], presenting considerable challenges in accurately phenotyping SCD pain. Consequently, this has limited our understanding of the etiology of chronic SCD pain, and has hindered the development of personalized treatment strategies. Gaining a better understanding of pain variability in SCD can inform more effective personalized approaches to managing chronic SCD pain. Similar approaches have been valuable in other chronic pain disorders, suggesting potential strategies for SCD [9–11].
Prior studies have used cluster analysis to identify patterns in pain symptoms and treatment response in adults suffering from various chronic pain disorders such as fibromyalgia, chronic low back pain, and temporomandibular disorder [9–16]. However, in SCD, attempts to identify homogenous pain subgroups using cluster analysis or similar machine learning methods have been limited [6,7,17]. Prior studies have shown that higher pain intensity and frequency are associated with worse psychosocial symptoms and lower quality of life. However, these studies included relatively small to moderate sample sizes, ranging from 62 to 291 participants, and only one study included participants from multiple centers. Moreover, only one of the three studies specifically examined individuals with chronic SCD pain, and none of the three studies compared the relationships between pain impact and both psychosocial and biological markers of disease severity.
Given the small number of studies, our understanding of interindividual variability in SCD pain remains limited. To address this gap, the primary objective of this current study was to identify distinct pain subgroups – both with and without chronic pain – by conducting a cluster analysis of a heterogeneous sample of adults with SCD enrolled in the Globin Research Network for Data and Discovery (GRNDaD) registry. We drew on findings from our previous work, which examined biopsychosocial variables associated with pain severity and pain impact among individuals with SCD, to select classification variables for the cluster analysis [3]. In that earlier work, we identified significant associations between psychosocial factors and both pain severity and pain impact – an SCD-specific measure of pain interference. For the current study, we hypothesized that while chronic pain would correlate with pain impact, differences in pain impact would emerge based on sex and biopsychosocial variables (e.g., total hemoglobin and emotional functioning), even within subgroups experiencing chronic pain. By identifying pain subgroups with shared demographic and biopsychosocial characteristics, we aim to deepen our understanding of pain variability among SCD participants with and without chronic pain, and to ultimately inform the development of more personalized and effective treatment strategies for chronic SCD pain.
Materials and methods
Study design and participants
This study is a cross-sectional analysis of data from the multi-institutional GRNDaD registry (https://grndad.org) as of April 2022. The registry enrolls and longitudinally follows participants with laboratory-confirmed diagnosis of SCD. Detailed eligibility criteria for participants in the GRNDaD registry have been previously published. Our analysis was limited to adults 18 years or older with the most common SCD genotypes – HbSS, HbSC, and HbS-beta thalassemia – who are enrolled in GRNDaD and whose baseline assessment include a response to the presence or absence of chronic pain. Fig 1 illustrates cohort selection for this analysis. Five hundred thirty-two adults from 11 sickle cell centers were included in this analysis. When these participants were compared to 612 adults who were not included in the study, no significant differences were found in their sociodemographic or clinical characteristics, except for exclusion of rare genotypes, as shown in S1 Table in S1 File.
Procedures
GRNDaD operates under a single institutional review board at Johns Hopkins University School of Medicine. Study personnel at each participating institution are responsible for obtaining written consent and enrolling participants. Enrollment for GRNDAD began January 2015 and remains ongoing. Data are securely stored in the Research Electronic Data Capture database (REDCap). Upon enrollment, participants complete validated self-reported questionnaires to assess pain, disease complications, and other patient-reported outcomes.
Measures
Sociodemographic measures.
Age, sex, education level, household income, and race were self-reported upon enrollment.
Disease-specific measures.
Laboratory values, including hemoglobin and fetal hemoglobin percentage (HbF%), were obtained from GRNDaD during the same calendar year as participant questionnaires were completed. In most cases, these lab values were collected within one month of completing questionnaires.
Pain-specific Measures. The primary outcome of interest is the Adult Sickle Cell Quality of Life Measurement Information System (ASCQ-Me) Pain Impact score [18], a 5-item scale that assesses pain interference over a 7-day recall period. This scale correlates with the Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference and Pain Behavioral questionnaires, and is considered a disease-specific measure of pain interference [19]. Raw scores were converted to T-scores using conversation tables in the ASCQ-Me User’s Manual [20]. The score range is 0–100, with higher T-scores indicating better health or less pain impact.
A baseline assessment of chronic pain is included with enrollment in GRNDaD. Specifically, participants are asked if they experienced pain on most days during the preceding 6 months. This approach is consistent with the published consensus definition of chronic SCD pain [21].
The frequency of SCD pain episodes was determined using item 1 from the ASCQ-Me Pain Episodes Frequency and Severity Measure subscale.
Psychosocial measures.
Two psychosocial constructs, social and emotional functioning, were assessed using the ASCQ-Me Social Functioning Impact Measure and Emotional Impact Measure, both 5-item scales. The social functioning impact measure evaluates participation in social activities and roles, while the emotional impact measure assesses psychological distress, and correlates with PROMIS depression and anxiety scales. Raw scores were converted to T-scores (range 0–100), with higher scores indicating better function in the specific construct.
Data analyses
In this study, we investigated pain variability between different individuals with SCD by first identifying and characterizing subgroups of GRNDaD participants using agglomerative hierarchical cluster analysis. We then examined the relationship between each identified cluster or subgroup and pain impact. Data were organized, cleaned, and analyzed using R statistical program (R Foundation for Statistical Computing, Vienna, Austria) version 4.3.2.
Our cluster analysis included 7 classification variables selected a priori based on evidence from the literature and our prior published analysis:[3] genotype (SS/Sβ0 or Variant [SC/Sβ+]), sex (male or female), baseline chronic pain (yes or no), age, maximum social and emotional functioning scores, and number of pain episodes (all from the same year as the worst pain impact score for individuals with repeated measures). Chronic pain was included as a classification variables because the clinical evidence indicated that variability in pain and pain-related outcomes is present even among SCD participants with chronic pain [5–7,17]. The rate of missingness was low for classification variables, and no missingness was found for genotype, age, sex, and baseline chronic pain. The remaining variables had a missingness of 10 percent or less, i.e., social functioning score (9.8%), emotional functioning score (10%), and number of pain episodes (9.8%).
We selected agglomerative hierarchical clustering due to its ability to provide an interpretable, tree-like structure that visualizes how individuals with similar pain characteristics are grouped at different levels of similarity. This method is particularly well-suited for our study because it does not require a pre-specified number of clusters, allowing for a data-driven determination of subgroup structure using the elbow plot and Gap statistic. Additionally, hierarchical clustering accommodates mixed data types (both continuous and categorical variables) when paired with Gower’s method, making it an appropriate choice given our classification variables. Ward’s minimum variance method was used as the linkage criterion to minimize within-cluster variance, ensuring compact and well-separated clusters.
We evaluated both an elbow plot and Gap statistic plot (S1 and S2 Figures in S1 File) to determine the appropriate number of clusters. Based on insight from both plots and an evaluation of the sample size of each cluster, we identified 6 distinct clusters.
Additional baseline and clinical variables of interest with missing rates ranging from 6%-33.8% were imputed after our cluster analysis, using the R package “mice” and our cluster variables (age, sex, genotype, chronic pain on enrollment, number of pain episodes, social, emotional and pain impact scores, and cluster number). We generated 5 imputed datasets, and pooled the results using Robin’s rules.
All baseline clinical and sociodemographic characteristics were summarized as median with interquartile range (IQR) and/or mean with standard deviation (SD) for continuous variables and as frequency with percentage for categorical variables. Differences between the 6 clusters were examined using Kruskal-Wallis rank sum tests and either Fisher Exact tests or Chi-Square tests for independence. Additional post-hoc comparisons were conducted for variables found to be significantly different between clusters. Finally, we conducted ordinal logistic regression to investigate the association between identified clusters and pain impact. We validated the model by assessing the proportional odds assumption. The reported results are odds ratios with 95% confidence intervals. To account for multiple comparisons, we adjusted the p-values using the Holm correction.
Results
Sample characteristics
The median age of the participants was 32 years (interquartile range [IQR] of 25–41 years), and 61% were female. Most participants (n = 370, 69.5%) had sickle cell anemia ([SCA], hemoglobin genotype HbSS or Sβ0), while approximately 31% had a variant genotype (hemoglobin SC or Sβ+). Chronic pain was reported by about 64% of the participants, and the median pain impact for the sample was 49.9 (IQR 44.4–58.0). Table 1 presents the baseline clinical and demographic characteristics of the sample.
Some notable differences were observed between females (n = 324) and males (n = 208) in the cohort. Males were younger, with a median age of 30 years (IQR 24–40), compared to females with a median age of 33 years (IQR 26–43), p = 0.016. Females also had lower median hemoglobin levels (7.9 g/dL) than males (10.2 g/dL), p < 0.001, and slightly lower social and emotional functioning scores than males, i.e., 51.5 vs 53.3, p = 0.016; and 50.5 vs 54, p = 0.008, respectively. Differences in chronic pain prevalence and other sociodemographic and clinical variables were not significant (Table 1). In addition to no significant difference in median pain impact scores among males (49.90 [44.40, 60.90]) and females (49.90 [44.40, 58.00]) (p = 0.470), we also noted no significant effect of sex on pain impact (OR = 0.91 [95% CI: 0.66, 1.27], p = 0.589). Further, in our model with sex, chronic pain and an interaction between the two, both sex (OR = 0.68 [95% CI: 0.37, 1.24], p = 0.207) and the interaction were insignificant (OR = 1.54 [95% CI: 0.74, 3.21], p = 0.248), though chronic pain was significant (OR = 11.11 [95% CI: 6.73, 18.33], p < 0.001).
Description and characteristics of clusters
Six distinct clusters, or pain subgroups, were identified – 3 with chronic pain (Clusters 1–3) and 3 without chronic pain (Clusters 4–6). Each classification variable differed significantly across the 6 clusters (Table 2).
Cluster 1: Female only, Chronic Pain, SCA – One hundred and thirty-four (25%) of the participants were included in Cluster 1. This cluster consisted solely of female participants with SCA, all reporting chronic pain. Half experienced 4 or more pain episodes in a 12-month period, and their social/emotional functioning scores were below average.
Cluster 2: 66% Female, Chronic Pain, Variant Genotype – One hundred sixteen participants (22%) comprised Cluster 2, which included only individuals with HbSC or HbSβ+ (variant genotypes) and chronic pain. Of these, 66% were female. This group was marked by frequent painful episodes and low social/emotional functioning scores.
Cluster 3: Male only, Chronic Pain, SCA – Ninety-two participants (17%) were grouped into Cluster 3. All were males with SCA and chronic pain. Social/emotional functioning scores were low.
Cluster 4: Male only, No Chronic Pain, SCA – Fifty-eight participants (11%) comprised Cluster 4, which was made up entirely of males with SCA and no chronic pain. Pain episodes were infrequent and social/emotional functioning scores were high.
Cluster 5: 59% Female, No Chronic Pain, Variant Genotype – Forty-six participants (8.6%) comprised Cluster 5. Approximately 59% were females. Chronic pain was absent in this cluster. Social/emotional functioning scores were high.
Cluster 6: Female only, No Chronic Pain, SCA – Eighty-six participants (16%) were included in Cluster 6. This cluster consisted solely of female participants with SCA and no chronic pain. Pain episodes were infrequent, and social/emotional functioning scores were high.
Comparison between pain clusters
Pain impact.
Distribution of median pain impact scores across the 6 clusters is depicted in Fig 2A. Most of the participants in the chronic pain subgroups (Clusters 1–3) had median pain impact scores below 50 (i.e., greater pain impact). Cluster 2 had the largest percentage of participants (76.9%) with greater pain impact, that is pain impact scores at or below 50. The subgroups without chronic pain (Clusters 4–6) had higher pain impact scores, indicating better health or less pain impact, and only a small subset of participants had scores at or below 50.
Pain impact assessed by ASCQ-Me Pain Impact Short Form (5-item questionnaire). Raw scores converted to T-scores. Higher T-scores imply better health or less pain impact. A. Plot depicting the distribution of pain impact T-scores for each cluster. B. Ordinal logistic regression modeling comparing the association of each cluster with pain impact.
We assessed the relationship between the identified clusters and pain impact. Among Clusters 4–6 (no chronic pain), Cluster 6 appeared to have greater odds of higher pain impact compared to 4 or 5; however, the differences between these clusters were not statistically significant. When corrected for multiple comparisons, no statistically significant differences in pain impact between the 3 chronic pain subgroups were found (Fig 2B). Cluster 2 appeared to have increased odds of having more severe pain impact (i.e., lower scores) compared to Cluster 1 (OR = 1.67 [95% CI: 1.03, 2.72]).
However, significant differences in pain impact were noted between each subgroup with no chronic pain and each subgroup with chronic pain. For instance, compared to Cluster 1 (female only, chronic pain, SCA), Cluster 6 (female only, no chronic pain, SCA) had significantly lower odds of experiencing more severe pain impact, with an odds ratio of 0.11 (95% CI: 0.06, 0.19, p < 0.001). This relationship was consistent in all comparisons between clusters without chronic pain vs clusters with chronic pain, with the clusters without chronic pain significantly less likely to report higher pain impact.
Additional variables.
We conducted detailed post-hoc comparisons of additional variables of interest (hydroxyurea use at the time of enrollment, admissions and emergency department visits in the 12 months preceding enrollment, education, income, total hemoglobin, and HbF%) to assess differences between the 6 clusters (S2 and S3 Tables in S1 File). Hydroxyurea use differed significantly between clusters with SCA (Clusters 1, 3, 4, and 6) and clusters with variant genotypes (Clusters 2 and 5), with approximately 80%-86% of participants with SCA on hydroxyurea. Similar patterns emerged when comparing hemoglobin levels, as median hemoglobin was significantly higher among participants in Clusters 2 and 5 (variant genotypes) vs participants in Clusters 1, 3, 4, or 6. No significant difference in hemoglobin levels was noted between Clusters 2 and 5. When examining HbF% levels, significant differences were noted only between Cluster 2 and Clusters 1, 3, 4, or 6. Compared to these 4 clusters, Cluster 2 had a significantly lower median HbF%. Percentage fetal hemoglobin, however, did not differ significantly between Clusters 2 and 5 (variant genotypes). No significant differences in HbF% were noted between Cluster 5 (variant genotype and no chronic pain) and Clusters 1, 3, 4, or 6.
Baseline admissions were significantly different between Clusters 1 and 6. Additionally, income varied significantly between Clusters 1 and 2, 1 and 5, 2 and 3, and 2 and 4 as well as 2 and 6, and 4 and 5. After adjusting for multiple comparisons, no significant differences in ED visits or education were observed between any 2 clusters.
Discussion
Here, we report the results of an analysis aimed at understanding variability in SCD pain experiences, using data from a multicenter SCD registry. Six distinct subgroups – 3 with chronic pain and 3 without chronic pain – were identified, highlighting the complexity and heterogeneity of pain experiences within the SCD population. As expected, subgroups with high baseline chronic pain (Clusters 1–3) differed significantly in pain impact compared to those without chronic pain (Clusters 4–6). While this finding aligns with our hypothesis and existing literature, the key significance of our study lies in its novel contribution to the SCD pain literature. Specifically, our study is one of the few to highlight the substantial burden of chronic pain in SCD through a detailed cluster or subgroup analysis, revealing that the presence of chronic pain predominates over other known associations with pain impact. These findings emphasize the critical importance of chronic pain prevention in SCD, particularly in the context of recent advances in disease-modifying and curative therapies. The results suggest that a greater focus on treating chronic pain may have a disproportionate impact on pain outcomes in SCD, and chronic pain prevention should be integrated as a central element in the overall care strategy for individuals with SCD.
Specifically, our analysis revealed that biological variables such as hemoglobin, SCD genotype, and HbF%, poorly accounts for the differences noted in high vs low pain impact subgroups, which extends the findings from our prior work [3]. For example, despite significant differences in biological variables (i.e., genotype, hemoglobin, and HbF%), Clusters 1 and 2 exhibited similar pain impact scores. Moreover, Clusters 2 and 5 shared similar biological profiles (i.e., same genotype and no significant differences in hemoglobin or HbF%), but differed significantly in pain profiles, primarily due to the presence or absence of chronic pain. Altogether, these findings suggest that chronic pain exerts a significant influence on pain outcomes in SCD, independent of traditional biological markers of disease severity – a relationship that is not well established in the context of SCD.
To the best of our knowledge, this is the largest study yet to use cluster analysis or machine learning methods to examine variability in pain experiences among individuals living with SCD. This study advances prior pain variability research in SCD [6,7,17] by using a larger multicenter cohort and emphasizes the pivotal role of chronic pain in driving pain outcomes, rather than relying solely on biological markers of disease severity. Additionally, our study addresses a gap in the literature by examining sex-related differences in pain impact within the context of chronic pain. Some prior studies have reported higher pain frequency and severity among females with SCD [8,22,23]. An analysis of 2,124 participants enrolled in the Sickle Cell Disease Implementation Consortium (SCDIC) registry, found that females with SCD were more likely to report higher pain frequency and pain severity [23]. A smaller, but longitudinal, study reported no sex-related differences in pain frequency or intensity [8]. Other studies have reported a greater prevalence of anxiety and depression among females with SCD compared to males, conditions often associated with worse pain [8,24,25]. However, these prior studies did not consider the presence or absence of chronic pain in their assessment of sex-related differences. Contrary to these studies, our analysis did not find significant differences in pain impact between male and female subgroups with similar genotypes and chronic pain prevalence. For example, Cluster 1, made up of females with SCA and chronic pain, and Cluster 3, comprised of males with SCA and chronic pain, had comparable pain impact scores. These observations were also present between Cluster 4 (females with SCA and no chronic pain) and Cluster 6 (males with SCA and no chronic pain). The lack of a significant effect of sex on pain impact was further confirmed by our regression models, which indicated that sex is not moderated by chronic pain. Our findings suggest that sex alone does not predict pain impact in our cohort of adults living with SCD, further emphasizing the need for personalized approaches to pain management that consider the interplay of multiple biopsychosocial factors.
Interestingly, we noted significant differences only in hospital admissions between Clusters 1 and 6, but no differences in ED visits between other clusters. However, this is consistent with findings from the Sil et al study [6], which examined variability in chronic pain in 62 children living with SCD. Differences in sociodemographic characteristics, specifically income and education, were variable in comparisons between clusters. A general pattern was observed in which participants in the clusters with a variant genotype had significantly higher income when compared to participants in clusters with SCA. This difference may be attributed to the difference in life expectancy and overall reduced risk for end-organ complications in individuals with a variant genotype when compared to those with SCA.
The key contribution of our study is the demonstration that chronic pain has an oversized influence on pain outcomes in SCD. Therefore, the prevention of chronic pain must accompany efforts at disease modification in SCD. Recently developed medications for SCD – L-glutamine, crizanlizumab, and voxelotor – have not consistently shown reductions in pain intensity or impact [26–30]. Over the past decade, transformative therapies for SCD, such as gene therapy and hematopoietic stem cell transplantation (HSCT), have been developed to eliminate or mitigate the effects of sickle hemoglobin by directly targeting the pathophysiology of SCD. The outcomes have been dramatic, with marked reductions in acute pain episodes and healthcare utilization, and improvements in quality of life, alongside improvements in biological parameters such as hemoglobin, hemolysis, and cerebral blood flow, a risk factor for strokes [31–34]. However, chronic pain post curative therapy persists in many individuals, with up to 40% and 18% still requiring opioids within the first and second years after treatment, respectively [32,35]. A few individuals continue to have episodic “vaso-occlusive” pain following these therapies. The impact of these therapies on chronic SCD pain, which is likely mediated by central mechanisms [36,37], remains to be fully understood. In addition to the unclear impact of curative therapy on chronic SCD pain, concerns such as donor availability, cost of therapy, and unknown long-term prognosis after gene therapy remain [26,38–41]. These limitations, in the context of our findings, do not diminish the tremendous impact of these advanced therapies for the individuals who receive them, but rather highlight the need for a better understanding and management of chronic pain in SCD. Our findings in the current study, combined with emerging research on outcomes following curative therapy, suggest that once chronic pain has developed, disease modification alone may not suffice to reduce pain severity and interference in many individuals living with SCD. Long-term pain outcomes for individuals who receive these therapies warrant further investigation, ideally through the systematic collection and analysis of longitudinal data from resources such as the GRNDaD registry. A deeper understanding of chronic pain and pain variability in SCD could facilitate the development of new and effective preventive pain interventions, and assist in identifying individuals who would benefit most from specific therapeutic interventions such as gene therapy.
While our study highlights the significant role of chronic pain in pain experiences among individuals with SCD, we were limited in the range of psychosocial variables we could examine. In a prior pain subgrouping study, Sil et al focused exclusively on pediatric patients with chronic SCD pain, and reported significant differences in psychosocial characteristics among the 3 pain clusters identified in their study. Psychosocial factors such as self-efficacy and pain coping strategies, are known to influence pain outcomes in other painful disorders [42]. While these factors are also believed to influence pain in SCD [4,43,44], their relative importance compared to other sociodemographic and biopsychological variables must be further investigated. Understanding protective factors by studying pain subgroups, such as Clusters 4 and 5, is critical to the design and implementation of preventive pain interventions for the SCD population.
Strengths and limitations
The interpretation of findings from this study should consider several limitations. The study was a secondary analysis of registry data, and thus, limits our ability to establish causality or predict pain outcomes. The reliance on self-report measures may introduce recall and nonresponse biases. While the classification and outcome variables used in analysis were informed by our prior research, they were limited to variables collected in GRNDaD, which may not fully capture the complexity of the pain experience in SCD. Additional subgroups or other clinically relevant distinctions between subgroups may have emerged if additional biopsychosocial measures known to influence pain outcomes had been included in the analysis. For example, incorporating specific pain measures such as number of pain locations or pain quality, may have revealed additional distinctions between the subgroups and potentially identified new subgroups. Furthermore, understanding the impact of early life experiences, such as childhood hospitalizations or behavioral health history could provide deeper insights into the development and persistence of chronic SCD pain [45]. Additional limitations include variations across SCD centers, especially regarding funding, resources, and the availability of non-pharmacologic pain treatments (e.g., acupuncture and cognitive behavioral therapy).
Conclusions & future directions
Our results here strongly indicate that the underlying cause of chronic pain must be addressed to significantly lower the burden of pain in SCD even as advancements in curative therapy continue. Preventing chronic pain must be a top priority aimed to improve quality of life for people living with SCD. Future research should prioritize our understanding of the underlying cause of chronic pain, and risk and protective factors for developing chronic pain in SCD. Additionally, further research is required to understand long-term pain outcomes in SCD in order to inform the development and refinement of preventive and symptomatic interventions for chronic SCD pain, and to identify individuals who would most likely benefit from emerging therapies.
Supporting information
S1 File. S1 FIGURE – Dendrogram from agglomerative hierarchical clustering using Ward’s method, illustrating the hierarchical structure of pain subgroups in individuals with sickle cell disease.
Colored boxes indicate the identified clusters at the selected cutoff level. S2 FIGURE – Assessment of agglomerative clustering solutions using (A) the within-cluster sum of squares (elbow method) and (B) the gap statistics. Dashed lines indicate the suggested optimal number of clusters based on each method. S1 TABLE – Demographic and clinical characteristics of the participant sample, including comparisons between included and excluded participants. S2 TABLE – Clinical and sociodemographic characteristics of the identified pain subgroups. S3 TABLE – Post-hoc comparisons of clinical and sociodemographic characteristics between pain subgroups.
https://doi.org/10.1371/journal.pone.0320889.s001
(PDF)
Acknowledgments
The authors thank all GRNDaD participants and staff for their commitment and investment in the GRNDaD registry.
References
- 1. Platt OS, Brambilla DJ, Rosse WF, Milner PF, Castro O, Steinberg MH, et al. Mortality in sickle cell disease. Life expectancy and risk factors for early death. N Engl J Med. 1994;330(23):1639–44. pmid:7993409
- 2. Smith WR, Penberthy LT, Bovbjerg VE, McClish DK, Roberts JD, Dahman B, et al. Daily assessment of pain in adults with sickle cell disease. Ann Intern Med. 2008;148(2):94–101. pmid:18195334
- 3. Kenney MO, Wilson S, Shah N, Bortsov A, Smith WR, Little J, et al. Biopsychosocial factors associated with pain and pain-related outcomes in adults and children with sickle cell disease: a multivariable analysis of the GRNDaD multicenter registry. J Pain. 2024;25(1):153–64. pmid:37544393
- 4. Matthie N, Jenerette C, Gibson A, Paul S, Higgins M, Krishnamurti L. Prevalence and predictors of chronic pain intensity and disability among adults with sickle cell disease. Health Psychol Open. 2020;7(1):2055102920917250. pmid:32426150
- 5. Sil S, Cohen LL, Dampier C. Psychosocial and functional outcomes in youth with chronic sickle cell pain. Clin J Pain. 2016;32(6):527–33. pmid:26379074
- 6. Sil S, Manikowski A, Schneider M, Cohen LL, Dampier C. Identifying chronic pain subgroups in pediatric sickle cell disease: a cluster-analytic approach. Clin J Pain. 2022;38(10):601–11. pmid:35997659
- 7. Bakshi N, Gillespie S, McClish D, McCracken C, Smith WR, Krishnamurti L. Intraindividual pain variability and phenotypes of pain in sickle cell disease: a secondary analysis from the Pain in Sickle Cell Epidemiology Study. Pain. 2022;163(6):1102–13. pmid:34538841
- 8. McClish DK, Levenson JL, Penberthy LT, Roseff SD, Bovbjerg VE, Roberts JD, et al. Gender differences in pain and healthcare utilization for adult sickle cell patients: the PiSCES Project. J Womens Health (Larchmt). 2006;15(2):146–54. pmid:16536678
- 9. Rabey M, Slater H, O’Sullivan P, Beales D, Smith A. Somatosensory nociceptive characteristics differentiate subgroups in people with chronic low back pain: a cluster analysis. Pain. 2015;156(10):1874–84. pmid:26020225
- 10. Vincent A, Hoskin TL, Whipple MO, Clauw DJ, Barton DL, Benzo RP, et al. OMERACT-based fibromyalgia symptom subgroups: an exploratory cluster analysis. Arthritis Res Ther. 2014;16(5):463. pmid:25318839
- 11. Wilson AT, Riley JL 3rd, Bishop MD, Beneciuk JM, Cruz-Almeida Y, Markut K, et al. Pain phenotyping and investigation of outcomes in physical therapy: an exploratory study in patients with low back pain. PLoS One. 2023;18(2):e0281517. pmid:36787322
- 12. Verra ML, Angst F, Brioschi R, Lehmann S, Keefe FJ, Staal JB, et al. Does Classification of Persons with Fibromyalgia into Multidimensional Pain Inventory Subgroups Detect Differences in Outcome after a Standard Chronic Pain Management Program? Pain Res Manag. 2009;14(6):445–53.
- 13. Abtroun L, Bunouf P, Gendreau RM, Vitton O. Is the efficacy of Milnacipran in Fibromyalgia Predictable? A data-mining analysis of baseline and outcome variables. Clin J Pain. 2016;32(5):435–40. pmid:26218005
- 14. Almeida SC, George SZ, Leite RDV, Oliveira AS, Chaves TC. Cluster subgroups based on overall pressure pain sensitivity and psychosocial factors in chronic musculoskeletal pain: differences in clinical outcomes. Physiother Theory Pract. 2019;35(12):1218–32. pmid:29771165
- 15. Alter BJ, Anderson NP, Gillman AG, Yin Q, Jeong J-H, Wasan AD. Hierarchical clustering by patient-reported pain distribution alone identifies distinct chronic pain subgroups differing by pain intensity, quality, and clinical outcomes. PLoS One. 2021;16(8):e0254862. pmid:34347793
- 16. Bair E, Gaynor S, Slade GD, Ohrbach R, Fillingim RB, Greenspan JD, et al. Identification of clusters of individuals relevant to temporomandibular disorders and other chronic pain conditions: the OPPERA study. Pain. 2016;157(6):1266–78. pmid:26928952
- 17. Knisely MR, Tanabe PJ, Yang Q, Masese R, Jiang M, Shah NR. Severe pain profiles and associated sociodemographic and clinical characteristics in individuals with sickle cell disease. Clin J Pain. 2021;37(9):669–77. pmid:34265788
- 18. Keller SD, Yang M, Treadwell MJ, Werner EM, Hassell KL. Patient reports of health outcome for adults living with sickle cell disease: development and testing of the ASCQ-Me item banks. Health Qual Life Outcomes. 2014;12:125. pmid:25146160
- 19. Keller S, Yang M, Treadwell MJ, Hassell KL. Sensitivity of alternative measures of functioning and wellbeing for adults with sickle cell disease: comparison of PROMIS® to ASCQ-Me℠. Health Qual Life Outcomes. 2017;15(1):117. pmid:28577358
- 20. Keller S, Yang M, Evensen C, Cowans T. ASCQ-Me User’s Manual. 2017.
- 21. Dampier C, Palermo TM, Darbari DS, Hassell K, Smith W, Zempsky W. AAPT diagnostic criteria for chronic sickle cell disease pain. J Pain. 2017;18(5):490–8. pmid:28065813
- 22. Barton-Gooden A, Grindley M, Knight-Madden J, Asnani M. Gender influences on the health of adolescents with sickle cell disease. Psychol Health Med. 2019;24(4):470–80. pmid:30318923
- 23. Masese RV, Bulgin D, Knisely MR, Preiss L, Stevenson E, Hankins JS, et al. Sex-based differences in the manifestations and complications of sickle cell disease: Report from the Sickle Cell Disease Implementation Consortium. PLoS One. 2021;16(10):e0258638. pmid:34714833
- 24. Levenson JL, McClish DK, Dahman BA, Bovbjerg VE, de A Citero V, Penberthy LT, et al. Depression and anxiety in adults with sickle cell disease: the PiSCES project. Psychosom Med. 2008;70(2):192–6. pmid:18158366
- 25. Sogutlu A, Levenson JL, McClish DK, Rosef SD, Smith WR. Somatic symptom burden in adults with sickle cell disease predicts pain, depression, anxiety, health care utilization, and quality of life: the PiSCES project. Psychosomatics. 2011;52(3):272–9. pmid:21565599
- 26. Brandow AM, Liem RI. Advances in the diagnosis and treatment of sickle cell disease. J Hematol Oncol. 2022;15(1):20. pmid:35241123
- 27. Vichinsky E, Hoppe CC, Ataga KI, Ware RE, Nduba V, El-Beshlawy A, et al. A phase 3 randomized Trial of Voxelotor in sickle cell disease. N Engl J Med. 2019;381(6):509–19. pmid:31199090
- 28. Niihara Y, Miller ST, Kanter J, Lanzkron S, Smith WR, Hsu LL, et al. A phase 3 trial of l-Glutamine in sickle cell disease. N Engl J Med. 2018;379(3):226–35. pmid:30021096
- 29. Ataga KI, Kutlar A, Kanter J, Liles D, Cancado R, Friedrisch J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429–39. pmid:27959701
- 30. Rai P, Ataga KI. Using disease-modifying therapies in sickle cell disease. Hematol Am Soc Hematol Educ Program. 2023;2023(1):519–31. pmid:38066905
- 31. Kanter J, Walters MC, Krishnamurti L, Mapara MY, Kwiatkowski JL, Rifkin-Zenenberg S, et al. Biologic and clinical efficacy of LentiGlobin for sickle cell disease. N Engl J Med. 2022;386(7):617–28. pmid:34898139
- 32. Leonard A, Furstenau D, Abraham A, Darbari DS, Nickel RS, Limerick E, et al. Reduction in vaso-occlusive events following stem cell transplantation in patients with sickle cell disease. Blood Adv. 2023;7(2):227–34. pmid:36240296
- 33. Hulbert ML, Fields ME, Guilliams KP, Bijlani P, Shenoy S, Fellah S, et al. Normalization of cerebral hemodynamics after hematopoietic stem cell transplant in children with sickle cell disease. Blood. 2023;141: 335–44.
- 34. Aumann MA, Richerson W, Song AK, Davis LT, Pruthi S, Davis S, et al. Cerebral hemodynamic changes after haploidentical hematopoietic stem cell transplant in adults with sickle cell disease. Blood Adv. 2024;8(3):608–19. pmid:37883803
- 35. Darbari DS, Liljencrantz J, Ikechi A, Martin S, Roderick MC, Fitzhugh CD, et al. Pain and opioid use after reversal of sickle cell disease following HLA-matched sibling haematopoietic stem cell transplant. Br J Haematol. 2019;184(4):690–3. pmid:29527656
- 36. Field JJ, Ballas SK, Campbell CM, Crosby LE, Dampier C, Darbari DS, et al. AAAPT diagnostic criteria for acute sickle cell disease pain. J Pain. 2019;20(7):746–59. pmid:30578848
- 37. Kenney MO, Knisely MR, McGill LS, Campbell C. Altered pain processing and sensitization in sickle cell disease: a scoping review of quantitative sensory testing findings. Pain Med. 2024;25(2):144–56. pmid:37815822
- 38. Hsieh MM, Fitzhugh CD, Weitzel RP, Link ME, Coles WA, Zhao X, et al. Nonmyeloablative HLA-matched sibling allogeneic hematopoietic stem cell transplantation for severe sickle cell phenotype. JAMA. 2014;312(1):48–56. pmid:25058217
- 39. Krishnamurti L. Hematopoietic cell transplantation for sickle cell disease. Front Pediatr. 2021;8:551170. pmid:33469520
- 40. Nikitin D, Beaudoin FL, Thokala P, McKenna A, Nhan E, Rind DM, et al. Gene therapies for sickle cell disease: effectiveness and value. J Manag Care Spec Pharm. 2023;29(11):1253–9. pmid:37889869
- 41. Kanter J, Thompson AA, Pierciey FJ Jr, Hsieh M, Uchida N, Leboulch P, et al. Lovo-cel gene therapy for sickle cell disease: treatment process evolution and outcomes in the initial groups of the HGB-206 study. Am J Hematol. 2023;98(1):11–22. pmid:36161320
- 42. Cohen SP, Vase L, Hooten WM. Chronic pain: an update on burden, best practices, and new advances. Lancet. 2021;397(10289):2082–97. pmid:34062143
- 43. Palermo TM, Lalloo C, Zhou C, Dampier C, Zempsky W, Badawy SM, et al. A cognitive-behavioral digital health intervention for sickle cell disease pain in adolescents: a randomized, controlled, multicenter trial. Pain. 2024;165(1):164–76. pmid:37733479
- 44. Matthie N, Jenerette C, McMillan S. Role of self-care in sickle cell disease. Pain Manag Nurs. 2015;16(3):257–66. pmid:25439112
- 45. Pernell B, Nagalapuram V, Lebensburger J, Lin CP, Baskin ML, Pachter LM. Adverse childhood experiences in children and adolescents with sickle cell disease: a retrospective cohort study. Pediatr Blood Cancer. 2022;69(3):e29494. pmid:34913574